G E. Youngblood et al./Composites Science and Technology 62(2002)1127-1139 either parallel or perpendicular to the fiber alignment direction. The error bars represent the range of mea- sured values at each temperature. As expected, the thermal diffusivity values measured for the parallel alignment were considerably higher than those for the perpendicular alignment because the interfacial gap resistances significantly interfered with heat conduction for the latter case Sic plug derived from Ceraset TM-Sic matrix, a PIP- To simulate the amorphous PIP pre-ceramic polymer 059…n was prepared using the composite fabrication procedure described earlier except without fibers. The plug con- tained numerous small isolated bubbles. but was crack free. Three thermal diffusivity samples, cut from the amorphous PIP-SiC plug, had an average bulk density Fig. 6. Thermal diffusivity values for the uniaxial Hi-NicalonMPI 1.95+0.04 g/cm. In Fig. 7, the calculated thermal con- the fiber direction (in air for T<400C. in argon for T>400C). The PIP-SiC matrix material are shown(lower curve).The error bars indicate the range of several measurements made at each thermal conductivity was only 0.720+0.04 W/(m K)at 27oC, and increased gradually up to 1. 10+0.03 W/(m K)at 400C. Since the PIP-SiC material was purposely The effective thermal conductivity values(Kefr)were made amorphous, the measured K-values were lower determined Iction of temperature(T) by calculating than K-values expected for fully crystallized PIP-SIC Km(T)=a(D)·p(D)·Cp(7 (8) material Also shown in Fig. 7 are the calculated thermal con ductivity values() for Hi-NicalonTMfiber where a(n) and p()are the measured thermal diffusiv- curve). To attain Kf, a series model was ity and bulk density values for each disk, respectively. where The bulk density at room temperature for each disk was determined from its mass and dimensions, where the Kerr= Vr(K+(1-VrKm (9) dimensions were adjusted for temperature dependence by using the thermal expansion of B-SiC. The specific a method suggested by Brennan et al. [8]. For the pur heat,Cp, was calculated using the rule-of-mixtures poses of this calculation, Vr=0.566, the Km-data were based on an equivalent composition of Hi-Nicalon in extrapolated up to 1000C(shown dashed in Fig. 7). terms of SiC, C and SiO2 phases and literature values of and Kefrvalues were calculated by Eq.( 8)using the Cp(T) for each[29]. At 323 K, the calculated Cp-value measured a(T)-data for the parallel alignment case for Hi-Nicalon was 689 J/(kg K), which compares well shown in Fig. 6. Because the Km - values were relatively with a measured Cp-value of 670 J/(kg K) listed by small, any uncertainty in their extrapolated values ippon Carbon Co [30] would make little difference in the Kr-calculation by Eq Two laser flash systems were used to measure the ( 9) thermal diffusivity by a standard method [31]. The The thermal conductivity of this batch of Hi-Nic thermal diffusivity tests were conducted between room lon fiber increased from about 4.6 W/(m K)at 27C temperature and 400C using a modified Anter Flash- up to 5.7 W/(m K)at 1000C. The average bulk density lineTM 5000(Flashline 5000 is a trademark used by determined for several representative Hi-Nicalon fiber Anter Corporation, Pittsburgh, PA, USA) laser bundles by the liquid gradient method was 2.69+0.02 g/ measurement system. From 400 to 1000 oC, a custom- cm. Nominal density and thermal conductivity values built system with a tungsten mesh tube furnace con- for Hi-Nicalon fiber are 2.74 g/cm and 7.8 W/(m K)at tained in a steel bell jar was used. The Anter system 25C and 10. 1 W/(m K) at 500C, respectively [30] operated in room air, while the custom-built system The lower bulk density and thermal conductivity contained a vacuum or inert gas. The temperature was measured for our particular batch of Hi-Nicalon varied typically by 100C steps in each system. To likely are due to batch fiber fabrication difference determine a(D), at least six measurements were made at also suggested by other work in our laboratory each temperature step. Further experimental details are To determine the effect of changing the atmosphere given elsewhere [32 on the transverse Keff, one Hi-Nicalon/PIP-SiC sample In Fig. 6, a(T)-data are plotted for the uniaxial Hi- with fibers aligned parallel to the flat specimen faces was Nicalon/PIP-SiC composite with the heat conduction mounted in the custom-built diffusivity system. TheThe effective thermal conductivity values (Keff) were determined as a function of temperature (T) by calculating Keff ð Þ¼ T ð Þ T ð Þ T CpðÞ ð T 8Þ where (T) and (T) are the measured thermal diffusiv￾ity and bulk density values for each disk,respectively. The bulk density at room temperature for each disk was determined from its mass and dimensions,where the dimensions were adjusted for temperature dependence by using the thermal expansion of b-SiC. The specific heat, Cp,was calculated using the rule-of-mixtures based on an equivalent composition of Hi-Nicalon in terms of SiC,C and SiO2 phases and literature values of Cp(T) for each [29]. At 323 K,the calculated Cp-value for Hi-Nicalon was 689 J/(kg K),which compares well with a measured Cp-value of 670 J/(kg K) listed by Nippon Carbon Co. [30]. Two laser flash systems were used to measure the thermal diffusivity by a standard method [31]. The thermal diffusivity tests were conducted between room temperature and 400
C using a modified Anter Flash￾lineTM 5000 (Flashline 5000 is a trademark used by Anter Corporation,Pittsburgh,PA,USA) laser measurement system. From 400 to 1000
C,a custom￾built system with a tungsten mesh tube furnace con￾tained in a steel bell jar was used. The Anter system operated in room air,while the custom-built system contained a vacuum or inert gas. The temperature was varied typically by 100
C steps in each system. To determine a(T),at least six measurements were made at each temperature step. Further experimental details are given elsewhere [32]. In Fig. 6, a(T)-data are plotted for the uniaxial Hi￾Nicalon/PIP-SiC composite with the heat conduction either parallel or perpendicular to the fiber alignment direction. The error bars represent the range of mea￾sured values at each temperature. As expected,the thermal diffusivity values measured for the parallel alignment were considerably higher than those for the perpendicular alignment because the interfacial gap resistances significantly interfered with heat conduction for the latter case. To simulate the amorphous PIP-SiC matrix,a PIP￾SiC plug derived from CerasetTM pre-ceramic polymer was prepared using the composite fabrication procedure described earlier except without fibers. The plug con￾tained numerous small,isolated bubbles,but was crack￾free. Three thermal diffusivity samples,cut from the amorphous PIP-SiC plug,had an average bulk density 1.950.04 g/cm3 . In Fig. 7,the calculated thermal con￾ductivity values for the simulated amorphous (1100
C) PIP-SiC matrix material are shown (lower curve). The thermal conductivity was only 0.7200.04 W/(m K) at 27
C,and increased gradually up to 1.100.03 W/(m K) at 400
C. Since the PIP-SiC material was purposely made amorphous,the measured K-values were lower than K-values expected for fully crystallized PIP-SiC material. Also shown in Fig. 7 are the calculated thermal con￾ductivity values (Kf) for Hi-NicalonTM fiber (upper curve). To attain Kf,a series model was assumed where: Keff ¼ Vf Kf  þ ð Þ 1Vf Km: ð9Þ a method suggested by Brennan et al. [8]. For the pur￾poses of this calculation, Vf=0.566,the Km-data were extrapolated up to 1000
C (shown dashed in Fig. 7), and Keff-values were calculated by Eq. (8) using the measured a(T)-data for the parallel alignment case shown in Fig. 6. Because the Km-values were relatively small,any uncertainty in their extrapolated values would make little difference in the Kf -calculation by Eq. (9). The thermal conductivity of this batch of Hi-Nica￾lonTM fiber increased from about 4.6 W/(m K) at 27
C up to 5.7 W/(m K) at 1000
C. The average bulk density determined for several representative Hi-Nicalon fiber bundles by the liquid gradient method was 2.690.02 g/ cm3 . Nominal density and thermal conductivity values for Hi-Nicalon fiber are 2.74 g/cm3 and 7.8 W/(m K) at 25
C and 10.1 W/(m K) at 500
C,respectively [30]. The lower bulk density and thermal conductivity values measured for our particular batch of Hi-Nicalon fiber likely are due to batch fiber fabrication differences,as also suggested by other work in our laboratory. To determine the effect of changing the atmosphere on the transverse Keff,one Hi-Nicalon/PIP-SiC sample with fibers aligned parallel to the flat specimen faces was mounted in the custom-built diffusivity system. The Fig. 6. Thermal diffusivity values for the uniaxial Hi-NicalonTM/PIP￾SiC composite shown in Fig. 5 measured parallel and perpendicular to the fiber direction (in air for T<400
C,in argon for T>400
C). The error bars indicate the range of several measurements made at each temperature. 1134 G.E. Youngbloodet al. / Composites Science andTechnology 62 (2002) 1127–1139